February 2025

Generative AI, the new revolution in technology, has already had a huge impact on many industries: marketing, legal, healthcare, software development, and many others. These tools started with basic use cases, like customizing templates or drafting basic messages, but now we are seeing the impact expand significantly, as AI can successfully perform complex tasks like web development or legal contract analysis in mere minutes.

These developments in AI present huge opportunities. Professionals who know how to use it can multiply their output and performance, while entrepreneurs can effortlessly automate complex tasks they would otherwise need to outsource. Many students have also started using AI for homework help or supplemental tutoring, and recent advancements have allowed specialized models to outperform human students in advanced high school math competitions (see this Scientific American article). So, for the next generation, knowing how to effectively utilize AI agents may become as important to being successful as knowing how to write an email.         

Inevitably, this technology and the debate on how to best handle it has already made its way to classrooms. Though we’re just scratching the surface, we have seen a lot of constructive and promising discussion through organizations like ISTE and NCTM on how to support students (and educators) in learning to properly and responsibly use this disruptive tech, without disrupting student learning and classroom dynamics. But there is still much work to be done, not just to understand the tools, but to ensure the policies and guidance available to educators from state, district, and school leadership match the needs of this technology. 

Is AI the new calculator? 

Discourse around AI in the classroom is often compared to the debate that arose when calculators became accessible to students in the 80’s. Would calculators be a gamechanger, allowing students to move past repetitive calculation and focus on problem solving and big picture math concepts, or would they leave kids unable to figure out a 20% tip on a $100 restaurant bill without reaching for a device? And have the skills themselves—the ones that students need to leave a classroom with to be successful—evolved as we contend with a world where most people have calculators (along with all the other features of a smartphone) in their pocket most of the time?   

The answer came down to teachers. Though calculator use in the classroom is still an active conversation, educators have adapted to teaching both how to use advanced calculator features for technical tasks and how to promote the underlying understanding of math so students leave classrooms with both the foundational knowledge and exposure to technology they might encounter. The struggle is between technology supporting learning and supplanting it.

So, is AI like a calculator? It is pretty close: on the surface, you ask it for the output you desire, but instead of a mathematical expression, you might request a recipe, a draft of an email, or an explanation of a concept. Recently, generative AI has gotten good enough at those things that it does feel calculator-like: you hit enter and the response feels pretty close to a “right answer.” But the key difference, which is critical to understanding what AI is doing, is that the underlying models are relational, pattern-based, and computational; in simple terms, they are designed to give you responses that are generated not via “thinking” in the human sense, but by giving you something that matches the pattern, i.e. “looks and sounds like” the answer. If you enter symbols into a calculator correctly, you will get a correct answer, guaranteed. If you ask AI a good question, you will probably get an answer that seems very reasonable, but figuring out whether or not it is actually correct can be a challenge. 

With that in mind, is what AI can do today impressive and mind-boggling? Absolutely. But you typically do not want to treat the output as a final result: AI can make mistakes (from simple errors to even sometimes citing nonexistent scientific papers by fake scientists to prove a point), and overall, it will produce texts that have a certain “mechanical” feel to them, that lack a human touch. So, AI-generated content should always be carefully reviewed, and possibly further investigated to verify its credibility.  

AI still certainly has many uses: it can help you brainstorm, it can draft, create outlines, let you bounce around ideas, break things down, revise/rephrase/edit your texts, and handle repetitive tasks— especially when you give it examples to mimic. This could mean using AI as a thought partner, using it for translation support, quickly generating datasets, or creating multiple versions of a problem (e.g. for retakes). But you always want to review the results and treat them as a starting point, not a final product. 

So while AI’s impact across education and many other fields may be even more wide-reaching than that of calculators, we are at a similar junction: we need to rely on teachers to set up the next generation for success so that AI becomes a useful tool for increasing the impact of one’s work, but not a substitute for thinking, logic, and foundational knowledge. 

What are some ways AI could be used in a math classroom? 

A Stanford study compared the output of teams that used AI to varying degrees in their work. They found two important facts: 

1. Teams that relied on AI to complete tasks typically underperformed compared to teams that did not use AI. 
2. Teams that used AI as a thought partner, going back and forth between AI and human team members to iterate on ideas and improve them, produced higher-quality output than both teams that didn’t use AI at all and teams that relied on AI to do the task by itself. 

By extension, teachers and students might also benefit from using AI as a thought partner. We suggest the following frames might be helpful: 

1. Rather than trying to replace instruction with AI, employ AI as a knowledge source for teams. 

AI can break down concepts and rephrase explanations to better reach all students, and it can patiently answer questions while you work with other teams.  

If you want to use AI as a thought partner, use it wisely:

• Ask it for 3 levels of explanation— for example, “Explain x to me like I’m a 2nd grader,” then “middle schooler,” then “expert.” 
• Ask it for additional examples or alternative explanations. 
• Ask it to compare and contrast two or more answers. 
• Ask it to refine an explanation so it is more clear. 
• Ask it to suggest more precise, mathematical language for your ideas. 
• Ask it to impersonate a historical mathematician or a professional for Q&A.
• Ask for real world examples to ground concepts, and to make them more relevant and approachable. 

When having students work with AI, consider subtly changing how you interact with teams. If you ask, “How tall is a stack of a million pennies?” then AI can tell you. If instead you say, “I heard someone stacked a million pennies and it reached the moon,” then human beings instantly have reactions while AI has trouble responding. Similarly, rather than prodding students to explain their reasoning, ask them to tell you the story of how they came up with their answer. These shifts in interactions will ensure that you’re hearing student thinking rather than an AI-generated estimate of a likely response…at least for now. 

2. Rather than trying to replace curriculum with AI, employ AI as a tutor. 

Here at CPM, we have been developing our own infinite practice system and we can tell you it is extremely difficult to get AI to generate anything remotely as good as what a human teacher, let alone a dedicated writing team, can produce. 

If you want to use AI as a tutor, lean into its strengths: 

• Tell it what you think you understand about a topic and ask it to give feedback.
• Have it ask you questions that interrogate your understanding.
• Ask it to provide edge cases with detailed explanations.
• Have it quiz you on common misconceptions and provide feedback to improve your responses. 
• Have it send you additional recall practice problems on a schedule, so you can spiral your new knowledge. 

3. Rather than trying to replace planning with AI, employ AI as a planning partner. AI is an expert at predicting what people might say. 

To maximize AI’s value as a planning partner, try the following: 

• Ask it to anticipate a range of student responses or give feedback on your anticipated responses. 
• Have it explain at least two valid approaches to solving a problem or give feedback on your explanations. 
• Ask it to identify how various solution approaches might connect or suggest additions to a list you provide. 
• Have it list the 3–5 most common misconceptions about a topic or review your list to ensure you didn’t miss anything. 
• Ask it for student-friendly metaphors that might help students understand key points or have it review your proposed metaphors for hidden biases.

4. Rather than trying to replace human connection with AI, employ AI to improve human communication: Use AI to communicate more effectively with more people more often. 

You might have AI: 

• Translate text or speech.
• Scan lessons for words or phrases that might cause confusion.
• Draft letters home about common topics.
• Revise emails/texts.

Note that while AI can rewrite texts at different levels, we should not let AI change the rigor of what students are engaged in. Use AI to identify, amplify, and explore parts of language that might cause confusion, rather than having it modify text for various students. 

5. Rather than spending valuable time on repetitive tasks, employ AI as a Teacher Aide: While AI can’t yet make copies, attend meetings, cut up cards for card sorts, or hand out supplies, it can quickly do other repetitive tasks. 

Try having AI do the following:  

• Populate a rubric table based on an outline. 
• Suggest fresh Door Questions.
• Generate new teams.
• Sort responses into categories.
• Review the gradebook for trends (but please only do this on AI local to your device so you don’t accidentally share private student data). 

As a note of caution, a recent Microsoft study found that, “[A] key irony of automation is that by mechanizing routine tasks and leaving exception-handling to the human user, you deprive the user of the routine opportunities to practice their judgment and strengthen their cognitive musculature, leaving them atrophied and unprepared when the exceptions do arise.” However, with how limited teachers’ time is, we may have to let that concern go to the wayside. 

6. Rather than trying to replace Professional Learning with AI, employ AI as an independent learning partner. 

One area AI can save a ton of time is reading and synthesizing research and academic papers. You can quickly extract key takeaways, interrogate papers to check your understanding, and identify areas for further learning. 

For instance, we’ve used a combination of Elicit, NotebookLM, and Claude to make sense of how researchers view particular topics. First, you can find papers on a topic of interest using Elicit. Then, plugging those papers into NotebookLM, you can request a study guide and ask questions about the paper. Finally, Claude can take all those notes and summarize them into a comprehensive, readable summary. 

An hour between those services allows digesting and synthesizing several lengthy academic papers, saving hours of time and accelerating learning, as well as helping identify the papers you might want to read in full yourself or areas for future exploration. 

When should technology be avoided?

Like all technology, AI certainly has its place, and also has plenty of areas where it can get in the way. So, when should you NOT be using AI and other technologies?

1. When it disconnects us: When students are collaborating, putting heads together over pencil and paper or a whiteboard (VNPS) is often way more powerful and frictionless without extra technology in the way. Though many technologies are meant to connect us over distances, when we can help it, there is no substitute for face-to-face collaboration. In a similar vein, as a teacher, you do not want tech to disconnect you from meaningful interactions and knowing each student. A conversation is more powerful than a digital note, tools like automatic grading make it easy to lose sight of what each student understands, and analytics dashboards always run the risk of flooding you with data to the point that you can lose the forest for the trees, or worse, of reducing students down to data points rather than humans. As much as technology has its place, nothing can replace the intuition, personal connections, and empathy of a teacher. 
2. When hands-on is better: In a time where almost everything you could do in a classroom has a digital equivalent, there is still very much a time and place for hands-on learning. For example, while graphing calculators are a powerful tool, graphing by hand also has its place. Technology needs to support human thought and creativity, not replace it. And while you may not be able to have students pilot a spaceship in class, the tactile feedback from algebra tiles or counting coins can be much more memorable and illuminating than staring at a screen.
3. When it is too much: With the plethora of tools available, especially after remote learning during the pandemic, it is easy to experience the illusion of variety when in reality students spent the whole class flipping between apps on their laptops. Digital fatigue is a real issue, and while a student may insist on taking notes with their device or playing that extra quiz game session, they will be much better off closing their devices now and then. 
4. When it is a roadblock: As a part of CPM’s development team, I (Misha) like to think our technology always works 100% of the time, and your experience always feels like sunshine, kittens, and unicorns. In reality, that is not always the case. You may have internet issues, you might stumble on an annoying bug (never on CPM’s systems, though!), or the tech just might not cooperate. We have seen teachers do some amazing pivots when tech that is supposed to be there… isn’t. It is totally okay to do something the tried-and-true way when it feels right to you. 

Parting Thoughts: Don’t Be Afraid to Experiment

AI has already proven to be a powerful tool and is certainly something that students will have to learn to live and work with during their academic and professional journeys. Many already use it. So, as if the job isn’t hard enough, teachers have to adjust to AI in and around their classrooms: an overwhelming task between students constantly learning new tricks on social media, districts trying to catch up with policies to technology that is evolving at break-neck speed, and a deluge of apps promising to make life easier doing this or that. You do not need much intelligence—artificial or otherwise—to see that that is a lot to deal with. But if there’s anyone who can do it, that’s teachers: don’t be afraid to experiment, break it down (make a list of ideas or tools you want to test out, then try one each week), and go back to tried-and-true methods when technology does not feel right. Most importantly, remember that no technology can ever replace the creativity, resourcefulness, intuition, human connection, and kindness of a teacher in the classroom. 

Want to Learn More?

For cool ideas and more resources on AI in the classroom, check out these resources: 

• ISTE’s collection of AI resources, lessons, and materials
• AI for Education free resource center (they also provide PD and paid courses)
• AI Educators downloadable resources (they also provide PD and paid courses)
• ControlAltAchieve AI Resources (articles, videos, and tools) 
• AI Educator Tools (a collection of AI tools for teachers) 
• AI Tools Club: a newsletter about trending AI tools (recommended for techies)

Authors’ note: Special thanks to Nicole Goerges, Lara Jasien, Nick Love, and Karen Wootton for their feedback and insights.